Research map / Vision-language and multimodal / Visual question answering
TransView: Inside, Outside, and Across the Cropping View Boundaries
Zhiyu Pan, Zhiguo Cao, Kewei Wang, Hao Lu, Weicai Zhong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add attraction modeling to content-aware region features (ablation) | TransView · small GAIC | Acc1/5: 62.0 -> 66.0 | Improved |
| add repulsion modeling to content-aware region features (ablation) | TransView · small GAIC | Acc1/5: 62.0 -> 64.5 | Improved |
| add repulsion modeling on top of attraction and content (ablation) | TransView · small GAIC | Acc1/5: 66.0 -> 68.5 | Improved |
| add attraction modeling on top of repulsion and content (ablation) | TransView · small GAIC | Acc1/5: 64.5 -> 68.5 | Improved |
| add content-aware RoI features to attraction and repulsion (ablation) | TransView · small GAIC | Acc1/5: 64.0 -> 68.5 | Improved |
| add learnable positional embeddings instead of none (ablation) | TransView · small GAIC | Acc1/5: 45.5 -> 68.5 | Improved |
| use learnable positional embedding instead of trigonometric encoding (ablation) | TransView · small GAIC | Acc1/5: 46.0 -> 68.5 | Improved |
| use learnable positional embedding instead of trigonometric encoding (ablation) | TransView · small GAIC | SRCC: 0.722 -> 0.803 | Improved |
| replace GAIC region-feature model with TransView | GAIC · GAIC test set | Acc1/10: 83.0 -> 85.4 | Improved |
| replace GAIC with TransView, both trained on GAIC | GAIC · FCDB | IoU: 0.673 -> 0.682 | Improved |
| include external attraction-aware words in the final feature fusion (ablation) | TransView · GAIC | performance | No clear effect |
| TransView trained on GAIC versus models trained without GAIC on FCDB | TransView · FCDB | IoU: 0.749 -> 0.682 | Hurt |
Closest papers by meaning
- Cropper: Vision-Language Model for Image Cropping through In-Context Learning (CVPR 2025 · similarity 0.87)
- Spatial-Semantic Collaborative Cropping for User Generated Content (AAAI 2024 · similarity 0.87)
- Learning Subject-Aware Cropping by Outpainting Professional Photos (AAAI 2024 · similarity 0.84)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.84)
- Venus: Benchmarking and Empowering Multimodal Large Language Models for Aesthetic Guidance and Cropping (CVPR 2026 · similarity 0.84)
- From Local Details to Global Context: Advancing Vision-Language Models with Attention-Based Selection (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
